design of a supply chain management dashboard …supply chain management (scm) is a key area that...
TRANSCRIPT
i
DESIGN OF A SUPPLY CHAIN MANAGEMENT
DASHBOARD
Cláudia Sofia Pedro Filipe
Information sharing to mitigate the bullwhip effect
Proposal for Project Work report presented as partial
requirement for obtaining the Master’s degree in
Information Management
i
ii
NOVA Information Management School
Instituto Superior de Estatística e Gestão de Informação
Universidade Nova de Lisboa
DESIGN OF A SUPPLY CHAIN MANAGEMENT DASHBOARD
INFORMATION SHARING TO MITIGATE THE BULLWHIP EFFECT
by
Cláudia Sofia Pedro Filipe
Proposal for Project Work report presented as partial requirement for obtaining the Master’s degree
in Information Management, with a specialization in Knowledge Management and Business
Intelligence
Advisor: Prof. Miguel de Castro Neto, Phd
November 2018
iii
ACKNOWLEDGMENTS
I would like to express my sincere gratitude to everyone that contributed to this project.
To my advisor, Professor Miguel de Castro Neto. This project would not have been possible without
his guidance, advices and constructive feedbacks.
To the company that was object of this project, for embracing it from the beginning and always provide
me resources and the most valuable and important data.
To my family, in particular to my mother. Thank you for the patience, understanding and for making
this challenge easier with your constant support and care.
To Marc, for always be there waiting for me with the most supportive and caring words. Thank you for
your motivation and for being my partner in all the challenges.
iv
ABSTRACT
Supply Chain Management is perceived as a key success area to organizations. It has developed
gradually through the last decades evolving from “the physical distribution to having a key role in
organizations’ competitiveness”.
The Bullwhip Effect is a major Supply Chain Management challenge, caused by the uncertainty and
variability in the demand that is amplified along the supply chain. This effect brings inefficiencies to
the supply chain in financial and operational terms.
The focus of this project is the exploitation of Company X’s Supply Chain Management data to develop
a dashboard that can display the overall information to the different actors evolved in this supply chain,
with the main goal of mitigate the bullwhip effect through information sharing.
This project starts with a literature review with two main focus: supply chain management and
business intelligence. It presents a framework of the supply chain management, how it is described
and its challenges. Besides that, it explores the best practices to develop effective information
visualization in a dashboard context.
A research was carried out in order to choose the best measures and metrics to be presented in the
dashboard, enhancing the use of the available data.
The output of this project is a dashboard displaying supply chain management measures and metrics
of demand forecast, inventory information, order information and purchasing forecast.
This project was designed and it is ready to be applied in a real context.
KEYWORDS
Supply Chain Management; Bullwhip Effect; Information Sharing; Data Visualization
v
INDEX
1. INTRODUCTION............................................................................................................. 1
1.1. BACKGROUND........................................................................................................ 1
1.2. PROBLEM DEFINITION ........................................................................................... 2
1.3. MAIN GOAL AND SPECIFIC OBJECTIVES................................................................. 3
1.4. STUDY RELEVANCE AND IMPORTANCE ................................................................. 4
1.5. METHODOLOGY ..................................................................................................... 4
2. LITERATURE REVIEW..................................................................................................... 6
2.1. SUPPLY CHAIN MANAGEMENT .............................................................................. 6
2.2. SUPPLY CHAIN MANAGEMENT AND BUSINESS INTELLIGENCE ............................. 8
2.3. PERFORMANCE MEASUREMENT ......................................................................... 11
2.4. INFORMATION VISUALIZATION ........................................................................... 12
2.5. DASHBOARD DESIGN ........................................................................................... 13
3. CONCEPTUAL MODEL PROPOSAL ............................................................................... 16
3.1. CONTEXT .............................................................................................................. 16
3.2. MEASURES AND METRICS.................................................................................... 17
3.2.1. DEMAND FORECAST ..................................................................................... 18
3.2.2. INVENTORY INFORMATION .......................................................................... 23
3.2.3. ORDER INFORMATION .................................................................................. 23
3.3. CONCEPTUAL MODEL .......................................................................................... 24
4. DEVELOPMENT ........................................................................................................... 28
4.1. PLATFORM ........................................................................................................... 28
4.2. DATA SOURCES .................................................................................................... 29
4.3. GRAPHIC AND DISPLAY RULES ............................................................................. 31
4.4. DASHBOARD ........................................................................................................ 32
5. RESULTS AND DISCUSSION ......................................................................................... 43
5.1. EVALUATION OF THE RESULTS ............................................................................ 43
5.2. PROJECT APPLICATION ........................................................................................ 44
6. CONCLUSION .............................................................................................................. 45
7. LIMITATIONS AND RECCOMENDATIONS FOR FUTURE WORKS ................................. 47
8. BIBLIOGRAPHY ............................................................................................................ 48
vi
LIST OF FIGURES
Figure 1 – Seasonality of Company X’s customer demand ........................................................ 3
Figure 2 – Overview of inventory depot/status along Company X’s SC..................................... 3
Figure 3 – DSR Process Model (DSR Cycle) (V. Vaishnavi & Kuechler, 2008) ............................. 5
Figure 4 – Sketch of a typical bullwhip model (Wang & Disney, 2016) ..................................... 7
Figure 5 – Prospective areas of future SC involvement (Daniels, 2017) .................................. 10
Figure 6 – Areas of focus for SC analytics (Partida, 2017)........................................................ 10
Figure 7 – The perfect metric (Juice, 2009) .............................................................................. 12
Figure 8 – The visualization process (Ware, 2013).................................................................. 13
Figure 9 - Reading Gravity (Pham, 2014).................................................................................. 14
Figure 10 – Chart Suggestions – A Tough-Starter (Abela, 2009) ............................................. 15
Figure 11 – SC of Group Y ......................................................................................................... 16
Figure 12 – SC of Company X.................................................................................................... 16
Figure 13 - Overview of inventory depot along Company X’s SC ............................................. 23
Figure 14 – Conceptual Model – Summary Module: Information Sharing to Mitigate the BWE
.......................................................................................................................................... 25
Figure 15 – Conceptual Model – Demand Forecast Accuracy Assessment Module ................ 25
Figure 16 – Conceptual Model – Future Demand Forecast Module ........................................ 26
Figure 17 – Conceptual Model – Inventory Information Module ............................................ 26
Figure 18 – Conceptual Model – Requested/Invoiced Orders ................................................. 27
Figure 19 – Conceptual Model – Purchasing Forecast ............................................................. 27
Figure 20 – Magic Quadrant for Analytics and BI Platforms (Howson et al., 2018) ................ 28
Figure 21 – Database Physical Model....................................................................................... 29
Figure 22 – Tables’ Relationships at Power BI.......................................................................... 30
Figure 23 – SCM Dashboard: Summary Module, Information Sharing to Mitigate the BWE, alias
“SCM Dashboard – Partilha de Informação para Mitigar o Efeito Chicote” .................... 34
Figure 24 – SCM Dashboard: Demand Forecast Accuracy Assessment Module, alias
“Monitorização de Previsão” ........................................................................................... 35
Figure 25 – SCM Dashboard: Future Demand Forecast Module, alias “Previsão de Venda” .. 36
Figure 26 – SCM Dashboard: Inventory Information Module, alias “Informação de Stock” ... 38
Figure 27 – SCM Dashboard: Requested/Invoiced Orders Module, alias “Encomendas” ....... 41
Figure 28 – SCM Dashboard: Purchasing Forecast Module, alias “Previsão de Compra” ....... 42
Figure 29 – Evaluation of the projects’ achievements ............................................................. 46
vii
LIST OF TABLES
Table 1 – BI benefits in the support of SCM challenges ........................................................... 11
Table 2 – Targeted users to the SCM dashboard ..................................................................... 17
Table 3 – Demand Forecast Accuracy Metrics ......................................................................... 19
Table 4 – Demand forecast metrics comparison, adapted (Kolassa, 2015) ............................. 21
Table 5 – Demand Forecast: Measures and Metrics ................................................................ 22
Table 6 – Inventory Information: Measures and Metrics ........................................................ 23
Table 7 – Order Information: Measures and Metrics .............................................................. 24
Table 8 – Entities and Attributes of the database.................................................................... 29
Table 9 – Summary of data required to measures and metrics............................................... 30
Table 10 – Adaptation from Juice’s Simple Font Framework (Juice, 2009) ............................. 31
Table 11 – Backgrounds definitions ......................................................................................... 31
Table 12 – Colours Chart .......................................................................................................... 32
Table 13 – Requirements to project application in other companies...................................... 44
viii
LIST OF ABBREVIATIONS AND ACRONYMS
APICS American Production and Inventory Control Society
APQC American Productivity & Quality Centre
BI Business Intelligence
BWE Bullwhip Effect
CFT Cosmetic, Fragrances and Toiletries
CSCMP Council of Supply Chain Management Professionals
DSR Design Science Research
IT Information Technology
KPI Key Performance Indicator
MAE Mean Absolute Error
MAPE Mean Absolute Percentage Error
OFR Order Fill Rate
SC Supply Chain
SCM Supply Chain Management
SKU Stock Keeping Unit
SLA Service Level Agreement
WAPE Weighted Average Percentage Error
3PL Third Party Logistics
1
1. INTRODUCTION
Supply Chain Management (SCM) is a key area that faces many challenges. This project focus on the
supply chain (SC) and specifically on the bullwhip effect (BWE) mitigation. To reach it, this project will
benefit from the use of information sharing and visualization, generating a great advantage to data
analysis, turning this key task efficient and intuitive.
This project main goal is the development of a BI model solution to monitor and control the SC. The
final delivery will be an SCM Dashboard aggregating relevant measures and metrics for information
sharing and decision making.
1.1. BACKGROUND
In the early 80s, Michael Porter introduced the concept of value chain as a different perspective of
looking to organizations. The main goal was to enhance customer value by adding it to products or
services. The starting point is to consider that a company is divided in two major group activities:
Primary activities and Support activities.
Primary activities are directly related with the creation, commercialization and support of a product or
service and include Logistics, Operations, Sales and Marketing, while Support activities assist the
operation of the Primary activities and include Human Resources, Procurement and R&D.
The value chain is presented as a valuable tool to identify where the great value lies in an organization
and how to increase it (Porter, 1985).
SCM exists in a fundamental exchange relationship and as a complement to value chain. While value
chain is providing the revenue that sustains supply chain, in parallel the supply chain is combining the
efforts to create and deliver products or services to the final customers (Cox, 1999).
SCM changed from the physical distribution to a “cost-effective, reliable and predictable service”.
(Arefin, Hoque, & Bao, 2015). This area is considered as a key role to stay competitive and to enhance
profit in challenging global markets (Childerhouse & Towill, 2003).
The area gained attention from organization’s strategic roles so it lead to the development of SCM. At
this point, the role of Information Technology (IT) has revealed as critical to the effectiveness of these
developments, taking advantage of the increase of computer power (Arefin et al., 2015).
The opportunity of using computer power for improvement of SCM is also a challenge, since it is
bringing more data and more efficient processes. To ensure these characteristics’ of excellence, the
need to measure and evaluate the processes rises.
As consequence, the need to acquire data and organize it has become fundamental. Business
Intelligence (BI) matches in this critical step, being defined as the whole process of managing data,
gathering it and transforming it into information that may support the decision making process
(Wieder & Ossimitz, 2015).
2
The analysis of data is supplied by Business Analytics through analysis models and data analysis
procedures. It is the use of data, statistical and quantitative analysis which “provide inputs to decision
making or even fully automated decisions” (Ittmann, 2015).
Davenport referred in ‘Competing on Analytics’1, that analytics focusing on SCM allows to “simulate
and optimize supply chain flows” and also “reduce inventory and stock-outs” (Davenport et al., 2006).
Ittmann also identified that the better tools to perform simpler analysis, and the usage of advanced
data visualization to present great volumes of data visually, are two focus of supply chain analytics
(Ittmann, 2015).
A picture is worth a thousand words, and information visualization is proving it. As a research field in
development, information visualization is being used to captivate users to analyse data in a more
effectively way. In a combination of data, eyes and mind, the information that could remain unnoticed
in tables or text, decreases when displayed by visualization (Few, 2008). The display of information
through visualization encourages the user to be alert, critic and to take actions.
1.2. PROBLEM DEFINITION
Group Y is a North American multinational company in the beauty industry which develops, produces
and sells cosmetics, fragrances and toiletries (CFT). The company operates with both own and
franchising operations in eighteen countries beyond the country of origin.
The company has been developing its activity in Portugal for more than ten years as Company X, the
first operation outside the origin country. With a multi-channel operation, the branch (with own
operation) oversees more than fifty stores in Portugal and an E-Commerce channel.
It was identified a BWE in the SC as there is high variability in demand forecast versus the actual sales.
Group Y and Company X have a two-echelon SC. Group Y is the only supplier of Company X, which
implies a huge dependence, and considering the geography of both organizations, a large lead-time
that can reach up to six months.
Company X manages its orders with an Order Up-To Model, also known as P model. At each month, a
replenishment order is submitted to Group Y, considering demand updates and on-hand/in-transit
inventory. In the same period a previous order should be received if the lead-time is respected.
The challenge is that Company X demand is non-smooth since the company products are a seasonal
choice for the final consumer, having peaks of sales that increase during festive seasons and
commemorative dates.
1 First time the term ‘analytics’ was used in an article title.
3
Figure 1 – Seasonality of Company X’s customer demand
SCM is considered, in parallel of Sales and Marketing, a key driver for business success. Nevertheless,
currently the company does not provide any SC information in a prompt and clear way.
Both Group Y and Company X systems are not integrated, meaning that all the products sold to
Company X are only recognized as inventory after being at the Portuguese distribution centre. Any
information about these products must be collected from offline support such as emails and excel files.
As result Company X does not have the knowledge of the total volume and investment in inventory
over the SC.
Besides the full track of the inventory, there is no sell-out or purchasing forecast available to the
interested departments. This is a clear opportunity to monitor demand forecast accuracy and improve
the process.
This lack of information creates a high dependency of different departments to the logistic and demand
planning departments, increasing the workload to the teams with manual and repetitive tasks of data
aggregation. In addition, this human intervention can easily drive to errors with high impact in the
operation. It also reduces time for logistics and demand planning departments to perform core role
activities like monitor and control the SC, identify problems and apply mitigation plans.
1.3. MAIN GOAL AND SPECIFIC OBJECTIVES
The main goal of this project is the design of a dashboard’s conceptual model for information sharing
to mitigate the BWE by focusing in the demand forecast, inventory and order information.
Figure 2 – Overview of inventory depot/status along Company X’s SC
To achieve the main goal, the following specific objectives have been defined:
1. Define relevant metrics to be monitored;
2. Assess the demand forecast accuracy and share future customer demand forecast;
Order RequestIn Transit to
Portugal (invoiced)
Distribution Center (3PL)
Point of Sales
4
3. Identify and share the inventory information along the SC;
4. Monitor order information and share purchasing forecast.
1.4. STUDY RELEVANCE AND IMPORTANCE
This project is justified by the following rational and impact in the company:
Local Branch and Group Wide: SCM is one of the core areas of the company that currently does not
beneficiates of the attention and investment needed to respond to the challenges proposed. It was
considered as an opportunity to focus on the BWE problem and to develop a project targeting the
mitigation of this problem. It was identified that sell-out information, inventory status and future
orders sharing mitigate the BWE. This is a lack in the company since this information is not being
shared. As a result, this is the major focus of this project.
Information Accuracy: nowadays all the reports are manual, through the compilation of data from
multiple sources. The information reliability is compromised so it is important to bring a solution that
can assure quality to the information shared.
Supply Chain: expected mitigation of BWE over the Company X and Group Y supply chains. By sharing
the sell-out demand forecast, inventory information and purchasing forecast, Group Y will have more
accurate information to plan, source and manufacturing activities.
Logistics Department: the department is bogged down in repetitive tasks, not having time to develop
analytical tasks and continuous improvement projects.
Demand Forecast: this process is not being assessed and has a major impact in the SC planning.
Without measuring it, the company doesn’t have the confidence to support the whole operation in
this data. It is considered critical to measure and share the results.
1.5. METHODOLOGY
The proposed methodology for this project implementation is the Design Science Research (DSR). This
methodology was chosen considering the main objective of the project: the creation of innovative IT
products or artifacts, useful and fundamental to understand and solve a question or a problem with
utility in the application environment (V. Vaishnavi & Kuechler, 2008), producing knowledge and
change.
In this project, the artifacts will be the dashboards and will be developed based in the knowledge flows,
process steps and outputs.
This approach presents five different steps of implementation: 1) Awareness of Problem, 2)
Suggestion, 3) Development, 4) Evaluation, and 5) Conclusion.
5
Figure 3 – DSR Process Model (DSR Cycle) (V. Vaishnavi & Kuechler, 2008)
Awareness of Problem: The knowledge, perception and concern of the problem. It was recognized a
BWE in the SC, being required to share information of demand forecast, inventory and orders over the
SC. This lack of knowledge undermines the welfare of the company by not knowing on time their risks
and opportunities. This project will be critical to move from a contingency management model to a
mitigation model with impact in operational departments (commercial and operational planning that
rule the operation directly) and in the general management (that will be allowed to criticalize and dig
in the data). The dashboard components’ will be defined by the output of the literature review with
focus in the BI in SCM (BWE). The awareness of the problem is explained in the Problem Definition
section.
Suggestion: Presentation of the tentative design of how the problem will be addressed. The
‘Suggestion’ is the creative step of the process, where the prototypes (artifacts) will be developed. The
design of the prototypes and the metrics will be chosen considering the best practices in the literature
review.
Development: Implementation of the tentative design. At this stage, the tentative design is completed
and developed. It will be necessary to define how to apply the designed solution. An analysis will be
developed to identify the best tool to implement the artifacts, meeting the company needs and the
available resources. The data that will feed the artifacts is property of the company and will be
provided to the selected tool.
Evaluation: Presentation of the performance measures. In this step, an assessment will be made to
identify the artifact implementation impact in the problem resolution. The evaluation may arise
awareness in other subjects related with the problem and bring new suggestions. As a result, new
developments and evaluation steps may concern possible adjustments.
Conclusion: Presentation of the research’s results. It will be determined if the artifact suited or not the
requirements for the problem resolution.
6
2. LITERATURE REVIEW
In this chapter is presented the literature review that served as theoretical fundamental to the
development of this project.
In first place, it is presented a description of SCM and its importance as a critical area for organizations.
The BWE explanation will be covered in this section.
The chapter follows with the interaction of the BI in the SCM area, and finally, it will be introduced
performance measurement and the best practices in information visualization and dashboards.
2.1. SUPPLY CHAIN MANAGEMENT
The term SCM was introduced in the early 1980s as an improved stage of logistics management,
looking outside the organization itself, and start focusing also in suppliers and customers including and
considering them in the overall process flow. (Haraburda & Col, 2016)
The importance of the SCM switched in the moment that it went from the physical distribution to the
insurance of a “cost-effective, reliable and predictable service”. (Arefin, Hoque, & Bao, 2015)
The Council of Supply Chain Management Professionals (CSCMP) defines SCM as the area that
“encompasses the planning and management of all activities involved in sourcing and procurement,
conversion, and all logistics management activities. Importantly, it also includes coordination and
collaboration with channel partners, which can be suppliers, intermediaries, third party service
providers, and customers. In essence, SCM integrates supply and demand management within and
across companies.”(Vitasek, 2013)
On the other hand, APICS (American Production and Inventory Control Society) defines it as “the
design, planning, execution, control, and monitoring of SC activities with the objective of creating net
value, building a competitive infrastructure, leveraging worldwide logistics, synchronizing supply and
demand, and measuring performance globally.” (Ariza, Buitrago, & R, 2008)
SCM evolved to a more complete area in an organization. Both CSCMP and APICS identify the following
as critical to properly define SCM: planning, monitor the SCM activities within its different links and
coordinate the supply and demand management.
It is demanding to have a good relationship with all the links involved – suppliers and customers – as
the SC represents the path of a product from the supplier until the final customer. There are many
steps before the landing product to the final consumer: they concern purchases, inventory control,
handling, storage and transportation. (Langlois & Chauvel, 2017)
SCM is a distinct area of organizations, assuming an important role to a company to stay competitive
and to enhance profit in challenging global markets (Childerhouse & Towill, 2003). SC costs may quash
from 50% to 70% of a company’s profit (Myerson, 2015), which shows a great opportunity to lower
costs through the SCM.
7
The cost reduction may touch different areas such as procurement, production, financial capital and
possession costs, transfer, breaking, product design and insurance costs. (Langlois & Chauvel, 2017)
For all of this, SCM presents itself as a key to figure out the continuous changing business environment.
Organizations are facing multiple challenges as the competition increases, the expansion of markets
arises, and the customers have more power and rising expectations. All of these challenges contribute
to a high rate of demand uncertainty (Simatupang & Sridharan, 2009).
The supply and demand uncertainty are key points of improvement. By working directly on it and in
the integration of activities, collaboration and information sharing through the different links of SCM,
arises the opportunity to reduce on-hand stocks, shorten throughput times and get a more competitive
SC (Millet, Schmitt, & Botta-Genoulaz, 2009).
The demand uncertainty causes the BWE, a phenomenon characterized by the variability in the
demand that is amplified along the SC, causing inefficiencies (Bray & Mendelson, 2012) as:
▪ high inventory;
▪ low service level;
▪ low quality;
▪ increased operation cost. (Jeong & Hong, 2017).
These inefficiencies are a consequence of the distortion of demand information that propagates
upstream through the different members of the SC causing mismatches such as the orders to the
suppliers tend to have a greater variance than sales.
These distortions will impact the interpretation of purchase orders, inducing in error manufactures by
the amplified demand patterns, leading to unplanned raw material purchases, hiring of resources to
accomplish the demand fulfilment and rising of manufacturing expenses. Other related costs impacting
all members propagate along the SC such as additional transportation and warehousing costs (Lee,
Padmanabhan, & Whang, 1997).
Figure 4 – Sketch of a typical bullwhip model (Wang & Disney, 2016)
The specific company of this project fits perfectly in a bullwhipped situation by its number of SC’s
echelon, order policy, lead-time and seasonal sales. Chen developed a study where it proves that BWE
8
exists in a two-echelon SC with the order-up-to inventory policy, increasing when the lead-time is
longer and forecast is less smooth (Chen, Drezner, Ryan, & Simchi-Levi, 2000).
A way to reduce BWE is by the decrease of demand distortion. The share of customer demand forecast
information among the members is one of the most effective ways in the BWE mitigation. (Jeong &
Hong, 2017) Besides customer demand forecast, also inventory and order information should be
available to upstream members of the SC with the same effect (Cannella, López-Campos, Dominguez,
Ashayeri, & Miranda, 2015).
As mentioned, one main critical element in the BWE cause is the unpredictability of the demand and
the need to forecast it, arising the challenge to choose the best forecasting method. Also, lead time is
a driving factor of demand amplification, which goes hand in hand with order policy. The information
sharing turns to be considered as an element to mitigate the BWE by sharing inventory levels, sales
data, sales forecast and order status (Wang & Disney, 2016).
2.2. SUPPLY CHAIN MANAGEMENT AND BUSINESS INTELLIGENCE
SCM gained prominence and attention from organizations’ strategic roles, leading to the development
of the area. The role of IT revealed to be critical to the effectiveness of these developments, taking
advantage of the increase of computer power (Arefin, Hoque, & Bao, 2015).
The use of computer power for improving SCM is an enormous opportunity because it is making
available more data, and “data now have the power to help business succeed”. For instance, retail
companies are dealing with huge volumes of data: “Wal-Mart handles more than a million customer
transactions each hour” (Ittmann, 2015).
Still, it is also a challenge, since more data does not mean more information. Like Wal-Mart, nowadays
companies are bogged down in data that continuously increases and it is mandatory to find a proper
way to understand it and gain competitiveness. This gain may come from the analysis of relevant
information, which desired output is how and when it may satisfy final customers (Langlois & Chauvel,
2017).
This take us to the need of having suitable tools and techniques to analyse and visualize data from
multiple sources in order to generate insights, which aim is to maximize cost reduction and minimize
the risks with direct impact in the operational agility (MHI; Deloitte, 2014).
Whereas in the past, data analysis was an area of competence which demanded specific technical skills
from employees that could extract data and come up with insights, nowadays BI is used to do so in a
quick, efficient way and on-time, providing reports and dashboards to support the decision-making
process (Langlois & Chauvel, 2017).
Langlois and Chauvel identified in 2017 that the main objectives for SCM as inventory reduction and
the full optimization of the SC cannot be possible without good information management, being
necessary to move from the traditional operational systems to rely in decision-making systems as BI,
concluding that “BI in management of SCM contributes to the differentiation of a business entity”
(Langlois & Chauvel, 2017).
9
Bi is defined as the whole process of managing data, gathering it and transforming it into information
that support the decision making process (Wieder & Ossimitz, 2015). BI is helping to “create knowledge
from a world of information. Get the right data, discover its power, and share the value, BI transforms
information into knowledge. BI is the application of putting the right information into the hands of the
right user at the right time to support the decision-making process” (Reinschmidt & Francoise, 2000).
The possibility of expansion of SC analytics is naturally rising as an opportunity opened by the following
events (Ittmann, 2015):
▪ Growing SC data;
▪ Cheaper data storage;
▪ Faster and ever-increasing processing power;
▪ Anywhere, anytime connectivity as mobile data is available almost everywhere;
▪ Better tools, making analysis simpler;
▪ Advanced visualization with tools and techniques to show and present huge volume of data
visually.
Ittmann elected in 2015 two major trends for SCM: big data and analytics, “the scientific process of
transforming data into insights for making better decisions”. Also referred that analytics is the use of
data, statistical and quantitative analysis to “provide inputs to decision making or even fully automated
decisions”. SC analytics focus is the usage of better tools to perform simpler analysis and the usage of
advanced data visualization to present great volumes of data visually.
Analytics focusing on SCM allows to “simulate and optimize SC flows” and also to “reduce inventory
and stock-outs” (Davenport et al., 2006).
Some companies started using BI tools to improve data visibility so as to reduce inventory levels and
also to better understand the sources of variability in customer demand to improve forecast accuracy
(Stefanovic & Stefanovic, 2009).
Furthermore, BI and analytics were identified as a top prospective areas of future involvement by
millennials working on SCM in a survey presented in figure 5 which was conducted by Peerless
Research Group in conjunction with SCM Review, APICS and APQC (American Productivity and Quality
Centre).
10
Figure 5 – Prospective areas of future SC involvement2 (Daniels, 2017)
APQC also launched a different survey where were identified the areas of focus for SC analytics (figure
6). Above 70% of the respondents consider cost reduction, data visualization, improve customer
satisfaction and productivity as areas of focus for SC analytics (Partida, 2017).
Figure 6 – Areas of focus for SC analytics (Partida, 2017)
Nevertheless, only a few studies are available on the interactions between SCM and BI, which raised
the need to relate the two concepts in order to identify its complementarity in a problem-solving
perspective:
2 676 respondents working in SCM with ages between 22 and 37
49%
45%
44%
41%
41%
40%
40%
37%
35%
30%
30%
29%
29%
28%
27%
Supply Chain design and planning
Business Intelligence/Analytics
Lean management (Six Sigma, etc.)
Demand planning, forecasting (S&OP, etc.)
Robotics and robotics process automation
Contracts and contract negotiations
e-commerce/e-fulfillment
Global trade management (import/export)
Purchasing/Procurement
Labor management
Engineering
Inventory management and solutions (i.e., WIP, etc.)
Manufacturing
Warehouse/DC management
Transportation and logistics
77%
75%
73%
70%
68%
65%
61%
Reduce cost
Provide visual tools like dashboards to helppeople in my organization obtain information
Improve customer satisfaction
Improve productivity
Provide more accurate forecasts
Reduce and mitigate risk
Contribute to supply chain optimization
11
Table 1 – BI benefits in the support of SCM challenges
SCM CHALLENGE BI BENEFIT
GROWING
SC DATA
“Today, as companies are drowned in
information which doubles every two to three
years, they have to find the best way to
understand it and gain a competitive
advantage.” (Langlois & Chauvel, 2017)
“Being able to consolidate and analyse this data
for better business decisions can often lead to a
competitive advantage, and learning to uncover
and leverage those advantages is what business
intelligence is all about.” (Stefanovic,
Majstorovic, & Stefanovic, 2006)
STOCK
REDUCTION
“Stock reduction and optimization of the supply
chain cannot be conceived without good
information management.” (Langlois & Chauvel,
2017)
“Beyond the traditional operational systems
that automate processes, it is without doubt
necessary to rely on an appropriate decision-
making system like BI.” (Langlois & Chauvel,
2017)
BWE “One (…) common phenomenon exhibited by
most SC systems is the BWE. (…) BWE is
generally viewed in extremely negative terms
because of its negative impacts on information
distortion, excess inventory levels, higher raw
material costs, overtime expenses, and added
shipping costs.” (Cannella, López-Campos,
Dominguez, Ashayeri, & Miranda, 2015) “It has
been advocated that the bullwhip effect can be
mitigated by information sharing.” (Wang &
Disney, 2016)
“(…) BI helps you to (…) get the right data,
discover its power, and share the value, BI
transforms information into knowledge. BI is the
application of putting the right information into
the hands of the right user at the right time to
support the decision-making process.”
(Reinschmidt & Francoise, 2000)
2.3. PERFORMANCE MEASUREMENT
The performance management is essential to a company, being one of the key success factors of
management in today’s competitive business environment.
Regarding the SCM, performance measurement provides how a company can assess whether its SC
has improved or degraded. “If a company does not have a clear understanding of how well its supply
chains are performing, it will be very hard to manage them successfully” (Stefanovic & Stefanovic,
2009).
The performance measurement plays a big role in an organization. According to Mahidhar, the
following five dimensions are the key of performance measurement (Mahidhar, 2005).
▪ Monitoring: measuring and recording actual performance;
▪ Control: identifying and attempting to close the gap between expected performance and
actual performance;
▪ Improvement: identifying critical improvement opportunities;
▪ Coordination: providing information for decision making and enabling internal communication
across processes as well as external communication with stakeholders;
12
▪ Motivation: encouraging behaviour and continuous improvement.
A metric can be defined as an observation that reduces uncertainty and it is expressed in a quantitative
form. (Hubbard, 2014) Being quantitative also provide the notion of progress when compared with the
objectives of the company. A measure should also be qualitative to provide understanding of the value
of the specific metric and strategic in order to be aligned with the company objectives (Nightingale,
2005).
With such importance in the performance measurement, a metric should be chosen considering that
it must be actionable, with easy interpretation, accessible and reliable for users, and transparent
(figure 7).
Figure 7 – The perfect metric (Juice, 2009)
Being actionable means that the problem is clear and it is possible to take actions whenever the metric
changes to values that requires that. It is advisable to focus on total measures rather than in changes
on those measurements.
The metric must be recognized and be transparent, using simple calculation that will be shared and
easy to understand. Create metrics by combining numerous measures will only bring confusion and
loss of engagement with the analysis.
The data that is leading to measures and metrics must be reliable and must come from sources that
users trust.
2.4. INFORMATION VISUALIZATION
A picture is worth a thousand words, and information visualization is proving it. As a research field in
development, information visualization is being used to captivate users to analyse data in a more
effectively way. In a combination of data, eyes and mind, information that could remain unnoticed (in
tables or text) decreases when displayed by visualization (Few, 2008). Displaying information through
visualization encourages the user to be alert and critic. As a result, it promotes call to action.
The visualization is an important part of the cognitive system, allowing us to acquire more information
through vision than from all the other senses combined, with multiple abilities:
13
▪ Visualization provides comprehension of large amount of data;
▪ Allows the perception of not anticipated data;
▪ Reveals errors of data treatment when applied in a suitable plot;
▪ Enhance the large-scale and small-scale features of data;
▪ Support hypothesis formation.
With this potential, visualization has become a very important part of the analysis and decision-making
(Ware, 2013).
Figure 8 – The visualization process (Ware, 2013)
Colour is a big ally of information visualization, playing an important role to ensure a soft presentation
of the display. “Colours must enhance data comprehension, not distract” (it is recommended to use a
maximum of six colours).
▪ Opposite colours should be used for comparison;
▪ Smaller values should be presented with less saturation and greater values with more
saturation;
▪ It is also recommended to use soft colours, except in a situation that the focus is to highlight
data. For backgrounds, neutral colours should be elected (Few, 2005).
2.5. DASHBOARD DESIGN
This project focus on the ability of BI to assist the analysis for the decision-making process – considering
the research study – of the SCM. The easy access to data analysis with proper metrics and performance
indicators to measure the SC are the aim of this project. For this reason, this section of the literature
review will focus on dashboard design guidelines.
14
In this project, the artifact to be developed is a dashboard. A dashboard, as a visual tool, arise to assist
and support people in the difficult task of understanding data.
Stephen Few identified in “Dashboard Confusion” in 2014 that a dashboard “is a visual display of the
most important information needed to achieve one or more objectives”. If a dashboard presents only
measurements it would be difficult to measure if the objectives are being achieved or not, how far we
are from them or how good performance we have. That is why metrics must be included in a
dashboard, because “what gets measured, gets managed” (Schmenner & Vollmann, 1994).
The information must be “consolidated and arranged on a single screen so the information can be
monitored at a glance” and the use of dashboards “to work faster and smarter” (Few, 2004) .
Few also identified in the same publication the attributes required to a dashboard to be effective:
▪ High-level summaries: effective communication, with information of what is happening but
not the root of the problem. The diagnostic may happen in the dashboard or not, since it is not
mandatory for a dashboard to drill-down into detail;
▪ Concise and intuitive display mechanisms: use of relevant and appropriate display
mechanisms according to the type of data and message;
▪ Customized: information must be presented to satisfy specific objectives defined for a person,
group or function;
▪ Metrics or Key Performance Indicators (KPI’s): sometimes a dashboard only presents
measurements (aggregation of data through operations as sum or average) but since it is
displaying information to achieve an objective, it should also monitor such performance.
Metrics and KPI’s should be included to compare achievements versus goals, but not in an
exclusive way since other information might be needed to complement the communication;
▪ Real-time information: only available in the dashboard if it is needed to achieve the objectives;
▪ Web Browsers: internet or intranet are the best available structures to present a dashboard,
though it is not mandatory to present it in such platform.
Figure 9 - Reading Gravity (Pham, 2014)
A dashboard must be well organized with information placement and size considering its importance:
important items appear larger and in strong reading areas (figure 9), and less important items appear
15
smaller. Related information should also be close and items that have ordered between them should
also be positioned in a way that can communicate that (Few, 2007).
As referred, one of the requirements for a dashboard to be effective is to have “concise and intuitive
display mechanisms”. Graphical representations are the elected artifacts to take advantage of the
users visual perceptions. Still, information should only be presented in this form if it brings more
perception to the user than presenting it on a table or text (Few, 2008).
Abela presents a chart suggestion considering the main objective of showing data and its specific
characteristics in order to have concise and intuitive display mechanisms (figure 10).
Figure 10 – Chart Suggestions – A Tough-Starter (Abela, 2009)
16
3. CONCEPTUAL MODEL PROPOSAL
In this chapter is presented the conceptual model proposal as the ‘Suggestion’ step of the DSR
methodology.
Initially it is made an introduction to the company context and process flows, followed by the measures
and metrics definition that will be based on a literature review. At the end of the chapter, the tentative
design is complete, and the conceptual model proposal is developed, being ready to be materialized
in the next chapter – “Development”.
3.1. CONTEXT
Considering the application of the project on the Company X’s SC, it is presented the SC of Group Y in
an overall perspective in figure 11. Group Y is responsible for the procurement and sourcing of raw
materials, as well as the production. Group Y detains all the factories (industry phase) and warehouses
to store products before selling them to branches.
The SC breaks down in the distribution phase, according to the products destination. If the destination
is the production country, the products are distributed by means of land until reach each retail unit.
On the other hand – and it is what concerns this project – the distribution of products that will be
exported is mainly made by maritime transport but, in case of emergency it can be transported by air.
Reaching the destination country, the product will be stored. The distribution makes it reach to the
retail units and to the final customer.
Figure 11 – SC of Group Y
In fact, Group Y and Company X are different companies. Company X depends entirely of Group Y since
it is its only supplier of products, which means that collaboration must be a key to this company’s
business. As such, Company X provides the forecasts of what will purchase to Group Y with a horizon
of eighteen months. This allows Group Y to plan the procurement, sourcing and manufacturing
activities. The SC of Company X is presented in the figure 12.
Figure 12 – SC of Company X
17
The supplier (Group Y) satisfies the Company X’s order request and the products are exported to the
destination country. Reaching the destination country, the product is centralized and distributed
according to the points of sale demand, where it will be available to final customers.
The development of this project will concern the Company X’s SC, with focus on demand forecast,
inventory information and order information, which were previously identified as critical to the
company.
3.2. MEASURES AND METRICS
As previous stated, a way to reduce the BWE is by decreasing the demand distortion. The share of
customer demand forecast information among the members is one of the most effective ways in the
BWE mitigation (Jeong & Hong, 2017). Besides customer demand forecast, also inventory and order
information should be available to upstream members of the SC with the same effect (Cannella, López-
Campos, Dominguez, Ashayeri, & Miranda, 2015).
The project will focus on dashboarding it and the target users are presented in the table 2.
▪ Demand Forecast: considering the major impact on Company X supplier to plan their SC and
the BWE that both Company X and Group Y are noticing, it was decided to assess the past
demand forecast and share future demand forecast information in order to decrease
distortion over the SC and measure results to manage the process;
▪ Inventory Information: it is critical to have knowledge of inventory status over the Company
X SC and the investment that it represents. It was also identified that the BWE has a major
impact in inventory. The inventory information analysis and sharing will allow to identify
variations in the inventory levels;
▪ Order Information: it was also identified as critical to share the forecast of purchases to
upstream members of the SC to mitigate the BWE. Since Company X only has a supplier, it
makes it even more critical, besides sharing, to assess past orders. A failure in orders can lead
to disruptions and delivery problems, affecting directly logistics, retail operation and planning
of activities. The project will present past order information and purchasing forecast.
Table 2 – Targeted users to the SCM dashboard
Theme Sub-theme Target Users
SCM
Summary
n.a. Operations’ Director, SCM Manager, Supplier
Demand
Forecast
Demand Forecast Accuracy
Assessment
Demand Planners
Future Demand Forecast Demand Planners, Marketing, Logistics, Supplier
Inventory
Information
Inventory Information Logistics, Supplier
18
Order
Information
Requested/Invoiced Orders Logistics, Supplier
Purchasing Forecast Supplier
For each metric it will be defined an alias for the original measure or metric to promote its
comprehension. The alias will be mostly in Portuguese language, since it is the official language of the
dashboard’s users.
The measures and metrics will be displayed for comparison over the historic time periods and
presented by different aggregations of product analysis by category and ABC analysis (the Pareto
principle: 20% of the products represent 80% of the sales). Any exception will be identified in the
specific section.
3.2.1. DEMAND FORECAST
Measuring demand forecast shows how well the Company X is predicting its upcoming demand. By
improving the prediction process, demand variation will decrease reducing the BWE.
The demand forecast is made in strategic and tactical levels. The time horizon of the demand forecast
is of 30 periods (approximately two years), which applies to the strategic forecast and it is used for
Business Planning of Company X but also to Group Y procurement, sourcing, manufacturing and
warehousing capacity planning.
On the other hand, the tactical demand forecast with a short-term horizon, serves the purpose of
planning the replenishment of point of sales by Company X.
The customer demand forecast will be presented and measured by campaign period. Each campaign
period has specific product launches (new product introduction), marketing actions and highlights.
CSCMP identified in the book “Suggested Minimum Supply Chain Benchmarking Standard” the
following as a good practice to measure forecast accuracy (CSCMP, 2010):
▪ Performance is measured against forecast;
▪ Historic is maintained for comparison purposes;
▪ Ability to measure forecast accuracy at the Stock Keeping Unit (SKU) level.
To define the demand forecast accuracy metrics, a literature review was made and it’s presented in
table 3.
19
Table 3 – Demand Forecast Accuracy Metrics
Author,
Year
Title MAE3 Mean Average Error
MAPE Mean Average Percent Error
WAPE4 Weighted Average Percentage
Error
MSE Mean Square Error
MASE Mean Absolute Scaled Error
(Kim & Kim,
2016)
A new
metric of
absolute
percentage
error for
intermittent
demand
forecasts
- "MAPE remains the preferred
method of business
forecasters and practitioners,
due to (…) its intuitive
interpretation. (…) It produces
infinite or undefined values
when the actual values are
zero or close to zero.”
“(…) the mean absolute error
can be scaled by the WAPE in
order to overcome the
problem of division by zero.”
- “The MASE is obtained by
scaling the forecast error
based on the in-sample mean
absolute error using the näive
(random walk) forecast
method, and can overcome
the problem of the MAPE
generating infinite or
undefined values.”
(Prestwich,
Rossi,
Armagan
Tarim, &
Hnich, 2014)
Mean-based
error
measures
for
intermittent
demand
forecasting
“Doubt has been cast on the
suitability of MAE for
intermittent demand”
“Percentage errors are
undefined if any sales = 0 and
have very skewed distributions
when sales ≈ 0. (…) Despite
these drawbacks MAPE is
recommended by most
textbooks”
- - “An advantage of MASE over
WAPE is that it is more reliable
on demand with seasonality,
trends or other forms of non-
stationarity. However, note
that the MASE of two series
with identical forecasts and
identical demands during the
forecast horizon will differ if
the two-series differed in their
historical demands. This is
counter-intuitive so MASE is
not always easy to interpret.”
3 MAE is also referred as MAD (Mean Average Deviation) in the literature. It will be defined as MAE in this project; 4 WAPE is also referred as Weighted MAPE, WMAPE, MAD/Mean Ratio, MAE/Mean Ratio or MMR in the literature. It will be defined as WAPE in this project.
20
Author,
Year
Title MAE Mean Average Error
MAPE Mean Average Percent Error
WAPE Weighted Average Percentage
Error
MSE Mean Square Error
MASE Mean Absolute Scaled Error
(Hoover,
2009)
How to track
forecast
accuracy to
guide
forecast
process
improvemen
t
“Overcomes many problems
with low-demand SKUs and
provides consistent measures
across SKUs.”
“MAPE is the primary forecast-
accuracy metric. Because is
scale-independent (since it is a
percentage error, it is unit
free), it can be used to assess
and compare accuracy across a
range of items. (…) MAPE is
also a very problematic metric
in certain situations, such as
intermittent demands”
“You want better accuracy
when forecasting those items
that, for whatever reason, are
more important than other
items. (…) Weighting permits
the forecaster to prioritize
efforts at forecast-accuracy
improvement”
-
“Compares the error from a
forecast model with the error
resulting from a naïve
method.”
(Hyndman,
2006)
Another look
at forecast-
accuracy
metrics for
intermittent
demand
"For assessing accuracy on a
single series, I prefer the MAE
because it is easiest to
understand and compute.
However it cannot be
compared between series
because it is scale dependent."
"Percentage error have the
advantage of being scale
independent, so they are
frequently used to compare
forecast performance
between data series. The most
commonly used metric is
MAPE."
“(…) scales the errors by the in-
sample mean of the series
instead of the in-sample mean
absolute error. This ratio also
renders the errors scale free
and is always finite unless all
historical data happen to be
zero.”
- “The MASE can be used to
compare forecast methods on
a single series, and, because it
is scale-free, to compare
forecast accuracy across
series. (…) The MASE is suitable
even when the data exhibit a
trend or a seasonal pattern.”
(Hyndman &
Koehler,
2006)
Another look
at measures
of forecast
accuracy
“If all series are on the same
scale, then MAE may be still be
preferred because it is simpler
to explain.”
“If all data are positive and
much greater than zero, the
MAPE may still be preferred
for reasons of simplicity.”
- “Historically, the RMSE (root
mean square error) and MSE
have been popular, largely
because of their theoretical
relevance in statistical
modelling. However, they are
more sensitive to outliers than
MAE or MdAE (median
absolute error), which has led
some authors to recommend
against their use in forecast
accuracy evaluation.”
“In situations where there are
very different scales including
data which are close to zero or
negative, we suggest the MASE
is the best available measure
of forecast accuracy.”
21
Other metrics that are more difficult to interpret were found during the literature review (Hyndman &
Koehler, 2006; Kim & Kim, 2016; Prestwich, Rossi, Armagan Tarim, & Hnich, 2014). It was decided not
to consider those since it is a concern to have clear metrics. If the problem and the metric are clear to
all users, it will be easy to take actions whenever the metric changes to values that doesn’t satisfy
company’s expectations.
Table 4 – Demand forecast metrics comparison, adapted (Kolassa, 2015)
Metric “Robust”? Defined? Comparable5? Intuitive?
MSE ꭗ √√ ꭗꭗ ꭗ
MAE √√ √√ ꭗ √
MAPE √ ((√)) √ √√
WAPE √ (√) √ √√
MASE √√ √ (√) ꭗ
Considering the characteristics of each metric identified in the tables 3 and 4, and the requirements
for a perfect metric identified in the ‘Performance Measurement’ section of the literature review, MAE
(Men Absolute Error), MAPE (Mean Absolute Percentage Error) and WAPE (Weighted Average Error)
were chosen as the metrics to assess the demand forecast accuracy.
▪ MAE – Mean Absolute Error
This metric measures the distance between the forecast and sales in absolute quantities, it is
a simple metric to understand (Hyndman & Koehler, 2006) and “provides consistent measures
across SKUs” (Hoover, 2009). When displayed at SKU level presents the specific error in
quantity of the SKU and when displayed in and aggregated level it shows an average of the
absolute error of observations.
𝑀𝐴𝐸 =1
𝑛∑|𝐴𝑡 −𝐹𝑡|
𝑛
𝑡=1
▪ MAPE – Mean Absolute Percentage Error
This metric measures the absolute deviation between forecast and sales in percentage.
In first place MAPE was chosen because of its simplicity and intuitive interpretation, with a
simple calculation, making it a clear metric (Hyndman & Koehler, 2006; Kim & Kim, 2016).
When displayed at SKU level presents the specific error in percentage of the SKU and when
displayed in and aggregated level it shows an average of absolute percentage error of the
observations.
MAPE is not advisable to be considered as a demand forecast accuracy metric when actual
sales are zero or close to zero (Hoover, 2009; Hyndman & Koehler, 2006; Kim & Kim, 2016).
5 Comparability between groups/series on different levels, e.g. fast vs slow selling products.
22
If the actual sales (𝐴𝑡 ) are zero the formula is undefined and in case of stock outs it may
produce high error values.
𝑀𝐴𝑃𝐸 =1
𝑛∑
|𝐴𝑡 − 𝐹𝑡|
𝐴𝑡
𝑛
𝑡=1
▪ WAPE – Weighted Average Percentage Error
This metric reflects the SKU importance through the assignment of weights according to the
most valuable business perspective to the analysis (Hoover, 2009). In this case the weights will
be assigned by revenue.
This metric solves the problem of undefined values when actual sales are zero or close to zero
(Hoover, 2009) since, instead of dividing the error by sales (that may assume zero in case of
stock outs or intermittent demand), it sums the errors and divides it by the sum of sales
avoiding undefined values.
MAE and MAPE are calculated both in SKU and aggregated levels, but WAPE is only calculate
at an aggregated level.
WAPE =∑ |𝐴𝑡 −𝐹𝑡 |𝑛𝑡=1
∑ 𝐴𝑡𝑛𝑡=1
Where:
𝐴𝑡=Actual value for t observations;
𝐹𝑡= Forecasted value for t observations;
t = number of observations.
In table 5 it is presented a summary of the measures and metrics that will be part of the Demand
Forecast section of the dashboard. This section has two different reports:
▪ Demand Forecast Accuracy Assessment – presenting past demands assessments;
▪ Future Demand Forecast – visualization of the upcoming demand forecasts.
Table 5 – Demand Forecast: Measures and Metrics
Report Alias Report Measure / Metric Description Alias Observations
Demand
Forecast
Accuracy
Assessment
Monitorização
de Previsão
Metric
MAE n.a. Measure by period
MAPE n.a. Measure by and over
periods WAPE n.a.
Measure
Past Demand Forecast Previsão
Actual Sales (quantity) Vendas
Actual Sales (revenue) Receita
Future
Demand
Forecast
Previsão de
Vendas
Measure Future Demand
Forecast
Previsão
23
3.2.2. INVENTORY INFORMATION
Inventory information was identified as critical to mitigate the BWE (Cannella, López-Campos,
Dominguez, Ashayeri, & Miranda, 2015) because one of the inefficiencies in the SC caused by the BWE
is specifically the increase of inventory ownership (Jeong & Hong, 2017).
It is considered critical to share the inventory status over the SC in this project. Inventory means
investment, especially for a company like Company X with a rigid SC, which works with long lead-time,
non-smooth customer demand and inventory depots in more than fifty points of sales beyond the
warehouse.
The knowledge of inventory volume allows to have a clear notion of the operation’s capacity. Valuating
the inventory as per cost of goods gives us an estimate of investment in inventory.
A clear picture of all inventory depots over the SC will be available in the artifact.
Figure 13 - Overview of inventory depot along Company X’s SC
Table 6 – Inventory Information: Measures and Metrics
Report Alias Report Measure / Metric Description Alias Observations
Inventory
information
Informação de
Stock
Measure Inventory Volume Volume Total units in stock
Inventory Investment Valor Total stock investment
3.2.3. ORDER INFORMATION
As third element of this project, it will be shared the order information. The share of order information
was also considered as a factor of BWE mitigation (Wang & Disney, 2016).
As mentioned before, the demand amplification causes the BWE and lead-time is a driving factor of
demand amplification (Wang & Disney, 2016). It impacts the Company X SC, making it rigid, which
demands a high capacity of inventory and operation planning to react to market with the inventory
resources available in destination country.
Lead-time is crucial in the order information since it reflects a clear understanding of the supplier
capacity at the time of the order placement (CSCMP, 2010).
Order RequestIn Transit to
Portugal (invoiced)
Distribution Center (3PL)
Stores
24
Table 7 – Order Information: Measures and Metrics
Report Alias
Report
Measure
/ Metric
Description Alias Observations
Requested/
Invoiced
Orders
Encomendas Measure
Number of
orders
Num.
Encomendas
Invoice volume Volume Units by order (confirmed)
Invoice cost Valor Cost by order (confirmed)
Lead-time Lead-time Time between order request and
distribution centre arrival
Days to ship Preparação Days from the order request to
shipment
Days in transit Trânsito Days in transit
Metric Order Fill Rate Order Fill Rate
(OFR)
The percentage of quantity met
according to client specifications
versus the total quantity
Purchasing
Forecast
Previsão de
Compra
Measure Forecast of
future orders
Previsão Volume forecast in future orders with
eighteen months horizon
3.3. CONCEPTUAL MODEL
Following the methodology, a conceptual model was developed as a suggestion of the final artifact.
Supported by the literature review, the conceptual model shows how to architect the dashboard.
At this stage, the development of the prototypes was determinant to ensure a coherent graphic display
of the information and to define how to combine the different visualizations.
This dashboard’s conceptual model is divided in six different modules, respecting the dashboard rules
and best practices identified in the literature review. The summary module compiles the most
important information of all themes and is presented in figure 17. The others five modules are focusing
on the different approaches of this project – demand forecast accuracy assessment (figure 18), future
demand forecast (figure 19), inventory information (figure 20), requested/invoiced orders (figure 21),
purchasing forecast (figure 22).
Following it is presented a generic structure of the dashboard to each module.
25
Figure 14 – Conceptual Model – Summary Module: Information Sharing to Mitigate the BWE
Instead of having only one module to present all information from demand forecast it was decided to
divide it in two modules: a module to assess the demand forecast accuracy and explore those
indicators with a past perspective and a module to present future demand forecast.
Although both modules have demand planners has target users, the Future Demand Forecast is a
module of interest to Marketing, Logistics and to the supplier so it was better to isolate it from Demand
Forecast Accuracy Assessment.
Figure 15 – Conceptual Model – Demand Forecast Accuracy Assessment Module
26
Figure 16 – Conceptual Model – Future Demand Forecast Module
Figure 17 – Conceptual Model – Inventory Information Module
Similar of what happened with the demand forecast, the same logic was applied in the order
information, having two modules: one for actual orders (requested, invoiced or delivered) and other
for the purchasing forecast (future orders).
27
Figure 18 – Conceptual Model – Requested/Invoiced Orders
Figure 19 – Conceptual Model – Purchasing Forecast
28
4. DEVELOPMENT
This chapter presents how to materialize the designed solution. In first place, an analysis will be
performed to identify the best tool to develop the artifact. Also, a comprehension of the data sources
will be presented followed by a graphic and a display rules manual. The chapter ends with the
presentation of the dashboard and analysis of the information displayed.
4.1. PLATFORM
The project will be developed using a Power BI platform from Microsoft. It was considered to be the
best solution for this project based in the Gartner Magic Quadrant for Analytics and BI Platforms
(Howson et al., 2018) (figure 20).
Since 2016 Microsoft was sharing the “leaders” quadrant with Tableau and Qlik. In that specific year
Microsoft was better in Completeness of Vision but was losing to the others in the Ability to Execute.
Figure 20 – Magic Quadrant for Analytics and BI Platforms (Howson et al., 2018)
During the last two years, Microsoft continued to be distinguished in the quadrant from the other
players in the Completeness of Vision and equated in a highest-level Tableau in the Ability to Execute.
29
Nowadays Microsoft has the highest composite scores for their Completeness of Vision and Ability to
Execute and a consistent position in the “leaders” quadrant, showing that Power BI has a differentiated
market position. This solution is a visionary product of a credible company and has high levels of
customer interest and adoption. It is also a competitive solution in terms of investment (Howson et
al., 2018).
4.2. DATA SOURCES
The data that will feed the artifacts is property of the Company X and will be provided to the Power BI.
Due to compliance issues and data protection, it was required to replicate the company’s database in
Microsoft Office Access to run the project proposal.
The entities and attributes of the database are presented in the table 8 and the physical database
model is displayed in the figure 21.
Table 8 – Entities and Attributes of the database
ENTITY ATTRIBUTE
PRODUCT Product_ID; Description; Category; Segment; Unit Cost; Inventory Quantity; Sales
quantity; Sales revenue; Sell-in forecast; Sell-out forecast; ABC.
INVOICE Invoice_ID (Inv); Transportation; Status; Order request quantity; Date of order
request; Date of shipment; Date of destination arrival.
DEPOT Depot_ID; Depot; Type; District; Municipality; City; Latitude; Longitude;
Figure 21 – Database Physical Model
The initial database was considered in the Power BI data model (figure 22), where calculated measures
were added to the model.
30
Figure 22 – Tables’ Relationships at Power BI
It is displayed in the following table the data required to produce the measures and metrics available
at the dashboard.
Table 9 – Summary of data required to measures and metrics
Data Table Measure/Metric Observation
Customer Demand Forecast T_Sell-out_Forecast MAE, MAPE,
WAPE, Future
Demand Forecast
Actual Sales quantity T_Sales MAE, MAPE,
WAPE
Actual Sales revenue T_Sales WAPE
Inventory quantity T_Inventory Inventory
information
Cost of Products T_Product Inventory
investment
Used to calculate the investment in
inventory
Order Request quantity T_Order_Request OFR
Invoice quantity T_Invoice OFR
Date of order request T_Invoice_Dates Lead-time
Date of shipment T_Invoice_Dates Lead-time
Date of destination arrival T_Invoice_Dates Lead-time
Forecast of future orders T_Invoice_Dates Lead-time
31
4.3. GRAPHIC AND DISPLAY RULES
The development of the prototypes will consider the importance of the information to define its
specific placement (Pham, 2014). Also, the information display will be chosen considering the chart
suggestions by Abela.
To guarantee the normalization across the different visualizations, the project will follow the following
graphic rules map, concerning font framework (table 9), backgrounds definition (table 10) and Colour
Charts (table 11).
Table 10 – Adaptation from Juice’s Simple Font Framework (Juice, 2009)
Purpose Size Font Colour Style
Body Clean readable text, 50-80% of
the text will look like this.
12 pts Verdana Black (Neutral)
#000000
Normal
No bold, no italic
1.2 line spacing
Header Separate and name major
sections of the writing.
Always start with upper case.
Left alignment.
18 pts
(150% of body)
Verdana Black (Neutral)
#000000
Bold
Whitespace above
Sub-Header Create a level of attention
between the Header and Body.
Always start with upper case.
Left alignment.
14 pts
(117% of body)
Verdana Black (Neutral)
#000000
Bold
Whitespace above
Notes Additional things a user should
be aware of, data sources,
metric calculations.
“Fade into the background”.
8 pts
(85% of body)
Verdana Grey
(De-
emphasized,
lower contrast)
#C8C8C8
Normal (no bold, no
italic).
Emphasis Draw the eye to key points that
need to make.
Same as body Verdana Orange (High
impact colour)
#FF5314
Bold
The Sub-Header format was added to the Graphic Rules Map to create a different level between the
Header and the Body. The format of Emphasis was not considered to this level because the main
objective is to identify a sub-theme, not to highlight data.
Table 11 – Backgrounds definitions
Colour Observation
Page #FFF
Transparency: 0%
For backgrounds, neutral colours should be elected. (Few,
2005)
Header #FFF
Transparency: 0%
Applicable for all modules but
“Dashboard” that has no fill.
Bar header positioned in the upper of each module.
Position X: 4 | Position Y: 1
Width: 1.271 | Height: 58
Applicable for all modules.
Charts #FFF
Transparency: 0%
No limits defined.
Tables #FFF
Transparency: 0%
No limits defined.
Table style: predefined.
32
Table 12 – Colours Chart
Number of data
identification
Colours by priority of user Observation
1 #01B8AA Columns Chart, Dispersion Chart
2 #01B8AA, #374649 (contrast for
comparison)
Line Chart & Columns, Pie Chart, Grouped Bar Cart,
Gauge
3 #01B8AA, #374649, #99E3DD Line Chart, Pie Chart, Map (according to depot type)
4 to 8 #015C55, #018A80, #34C6BB,
#67D4CC, #99E3DD, #E6E6E6,
#AFB5B6, #7F898A
Pie Chart, Grouped Columns Chart
More than 8 Use colours presented in the “4 to 8”
section and predefined on the 9th
colours onwards due to comparison
Ribbon Charts, Heat Map
Others Use predefined colours Ribbon Charts
4.4. DASHBOARD
The SCM Summary module (figure 23) presents an overview of the SC performance that allows the
user to identify where the useful information for his perimeter is and focus on the subjects of his
interest. This way of presenting information benefits the decision makers to get faster the information
needed to manage the operation by reducing the time for consultancy. This allows the user to be agile,
focusing their attention in the interest information and avoiding the waste of time.
All the modules are presented in Portuguese since it’s the official language of the target users.
The Demand Forecast Accuracy Assessment (alias “Monitorização de Previsão”) module (figure 24)
presents the three most important metrics to assess the demand forecast accuracy on a YTD (year to
date) basis: WAPE, MAPE and MAE.
The WAPE is of 0,52 when the target is 0,40 (company’s goal), which shows that there is space to
improve the process of demand forecast. When comparing the MAPE and WAPE, having a better WAPE
than MAPE indicates that the process of demand forecast is more accurate for items with more weight
in revenue, which indicates that the financial impact of the error is minor. That is obvious to the user
in the chart “WAPE | ABC” that shows that the error WAPE overall on A items is the smallest, followed
by the B items and C by last.
The cycle where MAPE and WAPE errors are the closest is 201808 with both metrics’ results improved,
hitting the best score on the YTD analysis. It is possible to see in the “WAPE | ABC” chart that all items
got more accurate demand forecast. Also, the MAE was the lowest in this cycle with 60 units (MAE
shows the units in average that the forecast is deviated from the actual sales).
The cycle with best results in terms of metrics is also the only cycle where the actual sales exceeded
the forecast, alerting the user to the fact that highest forecast is causing the error to increase.
Analysing the category’s performance, it is notorious that “Praia” and “Beleza” categories have the
highest WAPE (“WAPE | Categories” chart).
33
To better understand the impact of each category on the global WAPE it is advised to the user to
analyse the chart “MAPE | Receita por Categoria”.
Excluding “Acessórios”, “Praia” is the category with lowest revenue and by consequence, the lowest
impact in the global WAPE. The category “Beleza” has a higher revenue than “Praia” but it is below the
average. It is more important to have better MAPE in “Beleza” than in “Praia”. It is also easier to predict
the forecast of “Beleza” since “Praia” items are seasonal which increases the challenge.
What really stands out are “Corpo” and “Fragâncias” categories since they are the two categories with
highest revenue. “Fragâncias” is below but close to the average in WAPE so it is the category where it
should expend more effort to improve. It is also important to control and improve “Corpo” because
any fluctuation in MAPE may impact the global WAPE result due to its weight in the total revenue.
On the Future Demand Forecast (alias “Previsão de Venda”) module (figure 25) it is presented the
sell-out forecasted by cycle. The user is informed in the “Volume | YTG” data card of the volume
forecasted in the YTG period. The actual sell-out forecast is of 2,6 million units.
The other data card (“Volume | Seleção”) presents the sell-out forecasted to the selection that is
defined by the data segmentation “Categoria” and “Ano”. Besides the data card also the table and all
the charts are adjusted according to the data segments.
As seen in the previous module, “Corpo” and “Fragâncias” were the categories with highest revenue
and both have also the highest weight in the volume.
The cycle that has a higher forecast is 201815 because it corresponds to Christmas season and all the
promotions are already impacted. The forecast in 2019 and 2020 are smaller since they only reflect
the regular baseline sales. This is a point of alert because it will impact the purchasing forecast and, as
consequence, the industry planning.
When the user selects the year “2019” the forecast sell-out is 2,7 million units for the whole year.
Comparing with the 2,6 million units for the 2018 YTG appears to be unreliable but it is also a
consequence of not having the impact of campaigns in the sell-out forecast.
Analysing the segments, the user may take an insight that the sell-out forecast is aligned with the actual
sales since the three top segments weighting 38% of total forecast (“Loção Hidratante”, “Perfume EDT”
and “Creme Mãos e Pés”) belong to categories “Corpo” and “Fragâncias”, the two categories with more
weight in actual sales.
34
Figure 23 – SCM Dashboard: Summary Module, Information Sharing to Mitigate the BWE, alias “SCM Dashboard – Partilha de Informação para Mitigar o Efeito Chicote”
35
Figure 24 – SCM Dashboard: Demand Forecast Accuracy Assessment Module, alias “Monitorização de Previsão”
36
Figure 25 – SCM Dashboard: Future Demand Forecast Module, alias “Previsão de Venda”
37
The Inventory Information (alias “Informação de Stock”) module (figure 26) gives the user a clear
status of stock position. At this point, the Company X detains on-hand 3 million units (4,9 M€), divided
respectively in volume through the distribution centre and its points of sales depots (e-commerce and
physical stores) as 59,5%, 0,6% and 39,9%.
The company also detains 226K units that were already shipped and are in transit to the destination.
This stock represents and investment of 391K€.
The supplier was requested to sell 543,5 thousand units (849,7K€) but the orders are in request, which
means that may not be totally fulfilled.
During the first five months of 2018, the stock possession reached the highest volume when comparing
with the same period of the two previous years.
It is possible to observe that in the sixth month, the 2018’s volume remains stable when in the same
period of the previous year it was registered a considerable increase in the stock. This shows that even
with higher possession of stock, it is more stable than in previous years, suggesting an improvement
in the stock management.
Regarding the stock possession, it is a plus to analyse the “Posse Volume | Evolução YTD” chart to have
the knowledge where the stock is located. During the first three months of the year the stock level was
slightly lower in the distribution centre than it is in the four to six months of 2018. During the last three
months, the distribution centre and stores’ stock is more stable, suggesting that the stock is more
centralized and distributed to stores according to the sell-out needs.
Even though in the month six of 2017 the company possessed more stock in volume than in 2018, the
stock possession in value is the same (4,9M€). It is possible to conclude that the average cost price by
unit possessed is higher in 2018.
When analysing the stock possession by categories in volume it was expected to have “Corpo” and
“Fragâncias” as the two main categories since they were a trend in actual sales and sell-out forecast.
“Fragâncias” is in third place and “Beleza” is the second most possessed category. This category may
have problems of obsolescence in the future.
38
Figure 26 – SCM Dashboard: Inventory Information Module, alias “Informação de Stock”
39
The Requested/Invoiced Orders (alias “Encomendas”) module (figure 27) presents the information
about the orders that are already invoiced (delivered and in transit) and the ones in request from the
beginning of 2017 until the fifth month of 2018.
By selecting the “A” (actual) and “YTD” at “Período” data segmentation, the user compares the order
behaviour in the actual and previous year during the year to date.
In average, each order has 1.698 units and worth 2,80K€. By selecting the “Transporte” data segment
it is possible to analyse the differences between aerial and maritime. In average the maritime transport
has more units in the shipment than the aerial.
The regular transport is maritime which is used in 86% of the times. On the other hand, the aerial is
used in 14% of the orders.
In 2018 the company made 19 orders, when in the previous year made 38. It is making less orders in
this year but is also buying less, as it is possible to observe in the chart “Valor | Encomendas por Mês”.
The average lead-time is of 113 days and it does not diverge much from 2017 to 2018 (4 days).
It is notorious to conclude by analysing the “Preparação/Trânsito por Transporte” chart that the aerial
transport is increasing drastically the lead-time in the overall. In 2017 the supplier was taking 27 days
to prepare the order and ship it when in 2018 it is taking 87. This suggests that the supplier is losing
agility.
On the other hand, the transit days are shrinking from 41 to 20. Since the supplier is taking more days
to prepare the orders and the transit itself is the same (same flight hours), the handling to check in the
items in the distribution centre is being more efficient.
The maritime lead-time increased in three days which is residual. Nevertheless, the preparation days
of the supplier increased from 36 to 42, confirming the insight that the supplier is decreasing the
service level. Even though the service level is decreasing caused by the increase of preparation’s days,
the Order Fill Rate (OFR) is of 97,46%.
This may suggest that the Service Level Agreement (SLA) may not being respected. If the supplier – for
instance in case of shortage – prefers to increase the lead-time in order to guarantee the OFR when
what is more important to the Company X is to decrease the lead-time, there is a conflict of interests.
In this situation it is advisable to understand which KPI’s the supplier is being measured for.
Naturally if the problem is structural and operational it is necessary to guarantee that the supplier
works towards its solution and performance improvement.
The Purchasing Forecast (alias “Previsão de Compra”) module (figure 28) presents the information
about the future orders that are sent to the supplier each month. The Company X does not have any
commitment to acquire the specific quantity that forecasts but sharing it with the supplier allows the
supplier to better plan the industry and the order fulfilment process.
The user has eighteen months of purchasing forecast resulting in 4,3 million units and 7,3 M€. On the
2018’s YTG the months seven and eight are the ones with the highest volume of purchases,
representing the Christmas season replenishment. In 2019 only the eight month presents this peak.
40
This is the repercussion of only having the regular baseline sales in the sell-out forecast of 2019 which
impacts directly the industry planning. The supplier does not have a clear view of the volume and the
raw materials’ forecast won’t be enough to meet the needs of production. This will result in highest
cost of production due to low economies of scale in sourcing and by consequence a lowest margin.
41
Figure 27 – SCM Dashboard: Requested/Invoiced Orders Module, alias “Encomendas”
42
Figure 28 – SCM Dashboard: Purchasing Forecast Module, alias “Previsão de Compra”
43
5. RESULTS AND DISCUSSION
In this chapter will be presented the resulting SCM dashboard as output of this project and the
evaluation of the dashboard results.
The ‘Evaluation’ phase of the methodology evaluates if the project meets the expected output, or if it
requires more iterations to improve the current model.
5.1. EVALUATION OF THE RESULTS
The conceptual model of the SCM dashboard was conceived, designed and developed according to a
literature research to reach the best solution for the proposed problem.
Through the implemented data model, raw data is converted into reliable information that is related,
properly aggregated and released to the users through the dashboard in a web browser, allowing a
direct and easy access to data.
The interactivity offered by the dashboard is beneficial to the use and exploitation of data.
This dashboard is customized with information presented to satisfy the specific objective to produce
clear insights about the SC in a simple way to relate it with the BWE. It achieved the objective of assist
and support the users in the task of understanding data.
It is insured that users have full comprehension of the information presented. The measures and
metrics available in the dashboard are known by all the target users. Nevertheless, additional
explanations are provided in the dashboard to avoid misunderstandings.
Although most of the information presented are measures, the combination with metrics is crucial to
complement the communication to the user. Therefore, strategic metrics are also presented to
monitor the performance of the target achievements.
It is considered that the dashboard presents measures and metrics which reach the purpose of assess
the performance of the SC. It also promotes the generation of knowledge about the SC and give the
information needed to manage it.
The information display methods are relevant, concise and intuitive. The presentation is made
according to the type of data and the purpose of the analysis. The display methods provide the
comprehension of large amount of data and allow the perception of not anticipated data. Also,
enhances the large and small-scale features of data and support the hypothesis formation in the
analysis process.
This tool presents high-level summaries, which allows the user to adapt its analysis by data
segmentation, drill-down and drill-through in a simple and intuitive way.
Following the DSR methodology it is considered that the artifact implementation suits the
requirements for the problem resolution and as consequence other developments won’t be
performed.
44
5.2. PROJECT APPLICATION
This project was developed to Company X and Group Y but is applicable to any company with the same
characteristics than Company X.
The priority is to replicate this project in other branches of the Group Y due to the similarity of the
processes, data features and data sources.
Besides the Group Y’s branches, other companies are eligible to implement this project. Following, it
is presented the requirements to replicate the project entirely to other companies:
Table 13 – Requirements to project application in other companies
Characteristic Requirement Observation
Number of Suppliers One If a company has more than one supplier may implement the
project, but it won’t discriminate data from different suppliers.
It will treat all the supplier’s data as one.
Type of goods sold Products A company that sells services is not eligible
Period of demand forecast Cycle The cycle may assume the period the user demand.
The maximum number of cycles allowed are 30.
If the company forecasts in cycles of small periods it may be not
enough to the supplier consultation.
Order Up-To Model Monthly Purchasing is made once by month.
The monthly purchase can be made through multiple orders.
The purchasing forecast is made by month.
Order Status Requested
Invoiced
At the destination
Orders requested but not invoiced. Invoice is made before of
shipment.
Orders in transit that did not arrived at the destination.
Orders integrated in the on-hand inventory.
Number of Inventory
depots
Not defined The inventory depots are aggregated according to distribution
centre and type of sales’ channel
Product Master Two levels Product category and product segment.
ABC Defined to actual
sales
Only applicable to demand forecast accuracy assessment
45
6. CONCLUSION
The dashboard conceptual model is an IT artifact developed under the DSR methodology as a solution
to mitigate the Company X’s BWE over the SC. After verifying that the artifact suits the requirements
for the problem solution, we can conclude that the dashboard is potentially one of the best solutions
for organizing and present SC information and performance indicators.
This dashboard communicates the demand forecast, inventory information and order information.
These were the topics pointed in the literature review as critical to be shared over the SC links, which
have a positive impact in the BWE mitigation.
By compiling information of different SC areas, the developed dashboard has the potential to decrease
information distortion, to inform managers and SC’s users, to improve their knowledge and make them
more aware of SC as a whole, its activities and processes’ performance. As result, the users are
informed with clarity, the process of decision making is easier, more conscious and supported with
facts.
This dashboard is considered an SCM tool. Its contribution in the monitoring of the key BWE’s
offenders (demand forecast) and impacted factors (inventory, orders, lead-time) is critical to the SC
welfare.
By measuring the demand forecast accuracy, the process gets better managed and the improving is
expected. This leads to more accurate information to plan sourcing, manufacturing and operational
activities by the supplier. Better planning brings savings and cost reduction, reducing the product cost
and generating more margin to the Group.
By displaying the inventory information, the user has clear knowledge of the inventory status and
investment evolution over the SC. Also allows the company to monitor the unique supplier and require
service level standards.
The indicators play an important role in this dashboard, as such, its choice was based in the literature
review. It is considered that the indicators bring relevant insights to the analysis and to the decision
making-process. The measures and metrics available in the dashboard are known by the users, being
easy to make the dashboard available without complex training.
Beyond that, the data visualization enhances the potential analysis of the generated data, being a step-
forward development that improves the company’s perception of its SC, encouraging the user to be
alert and critic, promoting the call to action to improve the processes.
This evolution is a major step not only in information reliability but also for the analysts that are
currently responsible to generate this information. This change will have a positive impact in the work
satisfaction by allowing the analysts to focus on analytical tasks and continuous improvement projects.
The democratization of the SC information will allow other users to have access to the information.
Even though this project is targeting Company X, it shall be implemented in other Group Y’s branches,
providing an adapted and standard solution to each branch. The information democratization to Group
Y leads to better planning of sourcing and operation, driving to cost reduction.
46
This project can also be replicated to other companies with similar characteristics, being provided the
requirements to fit the implementation.
The following assessment (figure 29) presents the level of achievement of the:
• Initial goals proposed for this project;
• Degree of solution to the problems identified in the “Study Relevance and Importance”
section;
• Best practices of a dashboard creation identified in the literature review.
Figure 29 – Evaluation of the projects’ achievements
In the analysis of the final results, it is concluded that the artifact accomplish the goals proposed for
this project with success. The artifact is a step forward in the SCM of this company and group. SC
managers and users have the opportunity to follow the performance and improve processes with the
main goal of impact the decrease of the BWE.
0
2
4
6
8
10Presents the proposed dimensions
Decrease of information distortion
Decision making is easier,conscious and supported
User friendly
Improves perception of the SC
Generates knowledge of the SC
Promotes the call to actionInformation reliability
Information democratization
Encourages continuousimprovement
Use of data directly from thedatabase
Eliminate manual reporting
Have more reliable data
47
7. LIMITATIONS AND RECCOMENDATIONS FOR FUTURE WORKS
Following are presented the limitations occurred during the development of this project:
▪ Due to compliance issues and data protection it was not possible to access directly to the
company’s database;
▪ The lack of system integration between the Company X and Group Y did not allowed to
consider more relevant data to the SC dashboard;
▪ The process of gathering data was challenging since not all the data required is available at the
database. It was necessary to approach different departments to collect relevant data;
▪ The data collected from the departments had different formats and features. It was required
additional data treatment to achieve the quality desired to consider it in the project;
▪ The business characteristics and processes implemented make it impossible to analyse all data
with the same temporal periods (month or cycle). For this reason, it is not possible to apply
the same data segments or compare data with different temporal periods;
After the development of this project a few recommendations for future works are presented:
▪ Present real-time data by connecting the dashboard to the company’s database;
▪ Connect the Company X and Group Y databases to create synergies, improve data quality and
make it available and properly updated;
▪ Considering that not all the data is available at the database, the departments that own the
data should be responsible to provide it in proper format to the dashboard. This will arise the
departments’ sense of responsibility about the data displayed to the organization and lead to
an improvement of the data’s reliability;
▪ Targets to metrics and measures should be defined to provide a performance context and
guide the user during the analysis. Only “WAPE” and “OFR” present targets since are the
metrics with official goals;
▪ Measure the impact in the BWE of this dashboard’s information sharing through the evolution
of the measures and metrics;
▪ Consult the dashboard’s users to guarantee that the dashboard is a simple tool and meets the
business requirements;
▪ Apply a survey to measure the satisfaction about the dashboard in a defined period basis to
understand the users’ engagement.
48
8. BIBLIOGRAPHY
Abela, A. (2009). Chart Suggestions — A Thought-Starter, (\), 2009.
Arefin, M. S., Hoque, M. R., & Bao, Y. (2015). The impact of business intelligence on organization?s
effectiveness: an empirical study. Journal of Systems and Information Technology, 17(3), 263–285. https://doi.org/10.1108/JSIT-09-2014-0067
Ariza, S. G., Buitrago, S. G., & R, R. Y. R. (2008). APICS Dictionary. 13th Edition.
Bray, R. L., & Mendelson, H. (2012). Information Transmission and the Bullwhip Effect: An Empirical Investigation. Management Science, 58(5), 860–875. https://doi.org/10.1287/mnsc.1110.1467
Cannella, S., López-Campos, M., Dominguez, R., Ashayeri, J., & Miranda, P. A. (2015). A simulation model of a coordinated decentralized supply chain. International Transactions in Operational Research, 22(4), 735–756. https://doi.org/10.1111/itor.12175
Chen, F., Drezner, Z., Ryan, J. K., & Simchi-Levi, D. (2000). Quantifying the Bullwhip Effect in a Simple
Supply Chain_The impact of Forecasting, Lead-times and Information. Management Science , 46(3), 437–443.
Childerhouse, P., & Towill, D. R. (2003). Simplified material fow holds the key to supply chain
integration. The International Journal of Management Science, 31, 17–27. https://doi.org/10.1016/S0305-0483(02)00062-2
Cox, A. (1999). Power, value and supply chain management. Supply Chain Management: An International Journal, 4(4), 167–175. https://doi.org/10.1108/13598549910284480
CSCMP. (2010). CSCMP Suggested Minimum Supply Chain Benchmarking Standards.
Daniels, J., Tillman, J., Trebilcock, B., & Aschenbrand, J. (2017). Meet the not-so-average supply chain
millenials. Supply Chain Management Review, 48–55.
Davenport, T. H., Adams, L., Ahmad, Z. A., Karia, N., Anschutz, E. E., Becker, B., … Young, C. (2006). Competing on analytics. Harvard Business Review, 84(1), 98–107, 134.
https://doi.org/10.1177/2158244011433338
Few, S. (2004). Dashboard Confusion. Enterprise Magazine, 1–6. Retrieved from http://72.251.211.178/articles/visual_business_intelligence/dboard_confusion_revisited.pdf
Few, S. (2005). Uses and Misuses of Color. DM Review, 15(11), 62.
Few, S. (2007). Pervasive Hurdles to Effective Dashboard Design. Visual Business Intelligence Newsletter, (January), 1–7.
Few, S. (2008). A Menage a Trois of Data, Eyes and Mind. DM Review, 18(1), 14.
Haraburda, S., & Col, B. (2016). Transforming military support processes from logistics to supply chain management. Army Sustainment, 12–15. Retrieved from https://www.army.mil/article/162197/Transforming_military_support_processes_from_logistic
s_to_supply_chain_management/
Hoover, J. (2009). How To Track Forecast Accuracy To Guide Forecast. Foresight, summer(14), 17–24.
Howson, C., Sallam, R., Richardson, J., Tapadinhas, J., Idoine, C., & Woodward, A. (2018). Magic
49
Quadrant for Analytics and Business Intelligence Platforms. Retrieved March 31, 2018, from
https://www.gartner.com/doc/reprints?id=1-4PUK79W&ct=180130&st=sb&mkt_tok=eyJpIjoiWkRKa1lqWXpOR1E0Tm1RMiIsInQiOiJVYWhRd3E0Mk1FNERiSUplK2pCTjN5NXY2aDFtcTJrOGpIZ09ZSzluV1lqQTVRYzN2RStHV1B4c1J5N08rSUVZ
ajVhamlrSHBLMXJ0dHRhd2JWdmNcL09zaVVycHVodkwxSHJIYnRCSCtTN
Hubbard, D. W. (2014). How to Measure Anything: Finding the Value of Intangibles in Business. https://doi.org/10.1002/9781118983836
Hyndman, R. J. (2006). Another Look at Forecast Accuracy Metrics for Intermittent Demand. Foresight:
International Journal of Applied Forecasting, 4(June), 43–46. https://doi.org/10.1016/j.ijpe.2016.04.017
Hyndman, R. J., & Koehler, A. B. (2006). Another look at measures of forecast accuracy. International Journal of Forecasting, 22(4), 679–688. https://doi.org/10.1016/j.ijforecast.2006.03.001
Ittmann, H. W. (2015). The impact of big data and business analytics on supply chain management.
Journal of Transport and Supply Chain Management, 9(1). https://doi.org/10.4102/jtscm.v9i1.165
Jeong, K., & Hong, J.-D. (2017). The impact of information sharing on bullwhip effect reduction in a
supply chain. Journal of Intelligent Manufacturing. https://doi.org/10.1007/s10845-017-1354-y
Juice. (2009). A Guide to Creating Dashboards People Love to Use. Juice Analytics.
Kim, S., & Kim, H. (2016). A new metric of absolute percentage error for intermittent demand forecasts. International Journal of Forecasting, 32(3), 669–679. https://doi.org/10.1016/j.ijforecast.2015.12.003
Kolassa, S. (2015). Assessing predictive count data distributions.
Langlois, A., & Chauvel, B. (2017). The impact of supply chain management on business intelligence. Journal of Intelligence Studies in Business, 7(2), 51–61. https://doi.org/10.1108/13598540510606232
Lee, H. L., Padmanabhan, V., & Whang, S. (1997). Information Distortion in a Supply Chain: The
Bullwhip Effect. Management Science, 43(Llm), 668–673.
Mahidhar, V. (2005). Designing the Lean Enterprise Performance Measurement System. Engineering, 1–29.
MHI; Deloitte. (2014). The 2014 MHI Annual Industry Report: Accelerating change: How innovation is
driving digital.
Millet, P. A., Schmitt, P., & Botta-Genoulaz, V. (2009). The SCOR model for the alignment of business processes and information systems. Enterprise Information Systems, 3(4), 393–407. https://doi.org/10.1080/17517570903030833
Myerson, P. (2015). Supply Chain and Logistics Management Made Easy.
Nightingale, D. (2005). Metrics and Performance Measurement System for the Lean Enterprise .
Massachusetts Institute of Technology.
Partida, B. (2017). Maturity is key to analytics effectiveness. Supply Chain Management Review, (June), 56–59.
50
Pham, V. (2014). Best Practices in Data Visualizations.
Porter, M. E. (1985). Competitive Advantage:Creating and sustaining superior performance. New York
(Vol. 15). https://doi.org/10.1182/blood-2005-11-4354
Prestwich, S., Rossi, R., Armagan Tarim, S., & Hnich, B. (2014). Mean-based error measures for intermittent demand forecasting. International Journal of Production Research, 52(22), 6782–6791. https://doi.org/10.1080/00207543.2014.917771
Reinschmidt, J., & Francoise, A. (2000). Business Intelligence Certification Guide. IBM International
Technical, 166. Retrieved from http://capstone.geoffreyanderson.net/export/36/tags/proposal -1.0/proposal/research/sg245747.pdf%5Cnhttp://www.redbooks.ibm.com/redbooks/pdfs/sg24
5747.pdf
Schmenner, R. W., & Vollmann, T. E. (1994). Performance Measures: Gaps, False Alarms, and the “Usual Suspects.” International Journal of Operations & Production Management, 14(12), 58–69. https://doi.org/10.1108/01443579410072391
Simatupang, T. M., & Sridharan, R. (2009). Managerial Views of Supply Chain Collaboration. Gadjah
Mada International Journal of Business, 11(2), 253–273.
Stefanovic, N., Majstorovic, V., & Stefanovic, D. (2006). Supply Chain Business Intelligence Model. 13th CIRP INTERNATIONAL CONFERENCE ON LIFE CYCLE ENGINEERING, 613–618.
Stefanovic, N., & Stefanovic, D. (2009). Supply Chain Business Intelligence : Technologies , Issues and
Trends Introduction – Supply Chain. Ifip International Federation For Information Processing , 217–245. https://doi.org/10.1007/978-3-642-0
Vaishnavi, V., & Kuechler, W. (2008). Design Research in Information Systems. Order A Journal On The Theory Of Ordered Sets And Its Applications, 48(2), 133–140. https://doi.org/10.1007/978-1-
4419-5653-8
Vitasek, K. (2013). Supply chain management: Terms and Glossary. Healthcare Informatics : The Business Magazine for Information and Communication Systems, 17(2), 58–60.
https://doi.org/10.1201/9781420025705.ch2
Wang, X., & Disney, S. M. (2016). The bullwhip effect: Progress, trends and directions. European Journal of Operational Research, 250(3), 691–701. https://doi.org/10.1016/j.ejor.2015.07.022
Ware, C. (2013). Foundations for an Applied Science of Data Visualization. Information Visualization,
1–30. https://doi.org/10.1016/B978-0-12-381464-7.00001-6
Wieder, B., & Ossimitz, M.-L. (2015). The Impact of Business Intelligence on the Quality of Decision Making – A Mediation Model. Procedia Computer Science, 64, 1163–1171. https://doi.org/10.1016/j.procs.2015.08.599